You are on page 1of 29

Accepted Manuscript

Comparison of different calculation methods of Geological


Strength Index (GSI) in a specific underground construction site

Gábor Somodi, Ágnes Krupa, László Kovács, Balázs Vásárhelyi

PII: S0013-7952(18)30205-9
DOI: doi:10.1016/j.enggeo.2018.06.010
Reference: ENGEO 4867
To appear in: Engineering Geology
Received date: 2 February 2018
Revised date: 14 June 2018
Accepted date: 16 June 2018

Please cite this article as: Gábor Somodi, Ágnes Krupa, László Kovács, Balázs Vásárhelyi
, Comparison of different calculation methods of Geological Strength Index (GSI) in a
specific underground construction site. Engeo (2018), doi:10.1016/j.enggeo.2018.06.010

This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
ACCEPTED MANUSCRIPT

Comparison of different calculation methods of


Geological Strength Index (GSI) in a specific underground
construction site

Gábor Somodi, Ágnes Krupa, László Kovács, Balázs


Vásárhelyi
Gábor Somodi, 19. Esztergár L. str., Pécs, H-7633, RockStudy (Kőmérő) Ltd.,

PT
somodigabor@komero.hu
Ágnes Krupa, 19. Esztergár L. str., Pécs, H-7633, RockStudy (Kőmérő) Ltd.,

RI
krupaagnes@komero.hu
László Kovács, 19. Esztergár L. str., Pécs, H-7633, RockStudy (Kőmérő) Ltd.,

SC
kovacslaszlo@komero.hu
Balázs Vásárhelyi, 3. Műegyetem rkp., Budapest, H-1119, Department Engineering
Geology & Geotechnics, Budapest University of Technology and Economics, Budapest,
NU
Hungary, vasarhelyi.balazs@epito.bme.hu
MA
T ED
C EP
AC
ACCEPTED MANUSCRIPT

ABSTRACT

The Hoek-Brown (H-B) empirical failure envelope is widely used for calculating the
mechanical behavior of the rock mass. An important input parameter of the H-B failure
envelope is the Geological Strength Index (GSI), which was developed for the description of
rock structure and block surface conditions. Using the GSI, the mechanical behavior of the
rock mass can be calculated easily and accurately for both very poor and very good rock mass
qualities. Thus, knowing GSI value is an important input parameter for rock engineering

PT
calculations.

There are several qualitative methods which determine the GSI value accurately but no

RI
general international standard has been specified yet. Our objective in this paper is to analyze
the different published methods for the determination of the GSI based on observations during

SC
the construction phase of the Bátaapáti (Hungary) National Radioactive Waste Repository
(NRWR). Examinations of the values determined on-site gave significantly different results.
NU
Various correlations were determined between the calculated GSI values. Regarding the
different determination methods of the GSI, good correlation results were provided based on
MA

the rock mass classification data of tunnel construction phase.

It was found that there is no corresponding method for the determination of the GSI value
in all cases, the calculation of the GSI can be site-dependent. The applied rock mass
ED

classification process is recommendable for the further design and implementation works of
the NRWR and as an appropriate reference to other underground projects.
T
EP

keywords: Geological Strength Index (GSI), radioactive waste repository, rock mass
classification, granitic rock mass
C

Nomenclature
AC

GSI: Geological Strength Index


Ja joint alteration number (Barton et al, 1974)
Jc coefficient of joint conditions (Palmström, 1995)
Jn joint set number (Barton et al, 1974)
Jp Volumetric joint count (Palmström, 1995)
Jr joint roughness number (Barton et al, 1974)
Q: Rock Mass Quality (Barton et al, 1974)
RMi Rock Mass index (Palmström, 1995)
RMR: Rock Mass Rate (Bieniawski, 1989)
RQD: Rock Quality Designation (App. 2)
ACCEPTED MANUSCRIPT

Vb Block volume (Palmström, 1995)

PT
RI
SC
NU
MA
T ED
C EP
AC
ACCEPTED MANUSCRIPT

1 INTRODUCTION

During the design process in rock engineering, the Hoek-Brown failure envelope is used
mainly in brittle rocks for the determination of the rock mass failure envelope. In order to
determine the mechanical properties and parameters of the rock bodies, it is important to
describe their geology.

PT
One of the first attempts to involve a rock mass parameter in the boundary condition date
back to 1980 (Hoek & Brown 1980), but at that time it did not replace conventional rock mass

RI
classification methods. The first significant breakthrough was in 1995 (Hoek et al., 1995)
when it was defined that the quality of the rock mass depends on its structure and the joint

SC
surface quality, with a GSI between 10 (= extremely disturbed, very poor rock mass) and 100
(= intact rock). Since various numerical modeling methods have become increasingly
NU
prevalent at the time of the GSI, this parameter does not take into account the presence of
water, which is another input to modeling. It also ignores the strength of a solid rock, given
MA

that it is already included in the equation of Hoek-Brown failure envelope.

For more than two decades, GSI was used in several projects and rock engineering
problems. From the beginning, attempts were made to make the method suitable for
ED

classifying both very weak, fractured rock and very good, sound rock so that the mechanical
parameters of the intact rock can be converted to the Hoek-Brown boundary conditions. The
T

GSI value has been developed for heterogeneous rocks (i.e. flysch) and very weak rocks as
EP

well (refer to Hoek and Marinos 2000, Marinos 2014 and 2018).In the original definition of
the GSI, values were given by visual estimation – based on the well-known matrix by a
C

qualified, experienced engineer geologist (Hoek et al., 1995). From the beginning, there has
AC

been the need to create a clearer definition to provide accurate values. Up to now, several
significant, independent calculation methods have been published worldwide. They generally
tried to express the GSI by the application of existing rock mass classification methods used
in international practice. Morelli (2015) and Sari (2015) showed with Monte-Carlo methods
that theoretically significant differences were found between the different assay methods.

It has to be mentioned, that according to Marinos (2014), the behavior of the tunnel cannot
be predicted by GSI or any of the classification systems. The use of the classification ratings
must be accompanied by an understanding of the actual rock mass behavior in tunneling.
There has been a serious effort to develop guidelines and procedures for tunnel design in
ACCEPTED MANUSCRIPT

which the observation on the rock mass behavior is incorporated in the determination of
excavation and support classes (Schubert et al., 2003, Goricki et al., 2004).

Recently, Vásárhelyi et al. (2016) provided a brief analysis of the measured and calculated
GSI values, comparing the different qualitative methods in highly disturbed granitic rock in
Hungary. The GSI is also used for determining various mechanical parameters for the rock
mass (Vásárhelyi and Kovács, 2017). According to the results of Ván and Vásárhelyi (2014),
it is very important to know this value exactly. It was shown that sophisticated empirical

PT
equations can be highly sensitive to the uncertainties in the GSI values – even if the error of
the GSI is only 5 %, the relative sensitivity can reach 100 %.

RI
Deák et al. (2014) analyzed a large number of RMR, Q and GSI values from the tunnel
face surveys of Bátaapáti. According to their results, it is necessary to create location-specific

SC
correlations from one’s own database to avoid mistakes during design calculations. Their
results underline the importance of the proper determination of the GSI value. It is highly
NU
important for special tunnel systems such as the radioactive waste repository (Kovács &
Vásárhelyi, 2015). The aim of the present study is to continue previous scientific work by
MA

means of calculating and comparing GSI values with different methods from the rock mass
classification results of the construction phases of the Hungarian National Radioactive Waste
Repository. The large number of these evaluation can be used to analyze the discrepancy
ED

between different calculation methods. Results may help to choose an appropriate


determination method and draw attention to a few uncertainties of rock mass classification.
T
EP

1.1 Site description

The geographical position of the research area and the schematic view of the tunnel system
C

are shown in Fig. 1. The repository chambers are in 280 m depth. Up to now, a more than 6
AC

km long tunnel system has been excavated and all tunnel faces were documented (Deák et al.
2014, Kovács et al. 2015). This paper is focusing on rock mass classification data from the
four repository chambers and the two research tunnels.

Figure 1. a Schematic view of the National Radioactive Waste Repository (NRWR) at Bátaapáti, Hungary , b
tunnels reviewed in this paper (A: Access tunnels, B: research tunnel 3, C: Fault zone, D: Repository chambers),
c Repository chambers with the year of construction. Green: Monzonite and hybrid rocks. Purple: Monzogranite
rock. Red lines: Main shear zones with clay gauges. Green line: Basaltic dyke.

The facility was excavated in an intrusive and displaced Paleozoic granite batholith rock
body. The Mecsekalja Zone, an extended tectonic zone with complex geological and
ACCEPTED MANUSCRIPT

structural background, is located near this site. The granite is crosscut by NE-SW striking
trachyandesite dykes and even more frequently by randomly distributed aplitic veins. At
shallow depths the tunnels crossed a completely altered, weathered rock mass. Toward greater
depths, fractured but fresh granite is the common rock, sparsely intersected by a few meter-
thick fault zones. Carbonate, chlorite, hematite, limonite are the most typical fracture filling
minerals. Near the fault zones, high transmissivity damage zones occur but their fault cores
are characterized by intense clay mineralization that indicates low-grade hydrothermal
alteration and has very low transmissivity. Aside from the damage zones of faults, low water

PT
inflow rate or dry conditions are characteristic of the granite body.

RI
The “safety case” (determination of whether or not the proposed facility is able to safely
perform its function in time and space) of a potential deep geological repository for

SC
radioactive waste requires a sound understanding of both groundwater hydrogeology and
potential radionuclide transport through the rock mass (IAEA, 2003). The research area is in a
NU
fractured granitic body. Previous field observations and model results suggested that the rock
formation of the repository is strongly compartmented hydraulically, dividing the
MA

underground flow system into several blocks of limited hydraulic connection (Benedek et al.,
2009, Benedek and Molnár, 2013). Each block and the boundaries are characterized by
different fracture orientation as well. Based on field observations, these have been divided
ED

into the so-called more transmissive zones and the less transmissive blocks. The repository for
low- and medium level nuclear waste disposal is placed in a less transmissive hydraulic
T

compartment. Since the boundaries between the blocks are fault zones with clay gauges, it
EP

was important to study this natural barrier in detail.

From a geotechnical aspect, four main rock types can be distinguished in the granite
C

formation: monzogranite, monzonite, hybrid rocks and igneous veins. The four repository
AC

chambers were excavated in a rock mass which is mainly composed of monzogranite with
aplitic veins and scarce monzonite enclaves.

1.2 Rock mass characterization of the investigation site


During the excavation of access tunnels and the NRWR, the GSI, RMR and Q values were
determined for all tunnel faces (Deák et al., 2014, Kovács et al., 2015, Vásárhelyi et al.,
2016). Plan used five different rock mass classes (Class I…V) for determining the tunnel
support during excavation. Examples for the different types of rock masses are presented in
Figs. 2a-2d. In the construction design of the tunnel system, the behavior of the rock masses
ACCEPTED MANUSCRIPT

was specified by using the GSI system. In this paper, two short research tunnels and four
repository chambers were analyzed.

Although the underground tunnel system was excavated in a rock mass of fair quality, the
tunnel walls were supported by shotcrete and rock bolts after every blasting advance step. The
site geotechnical engineer was required to carry out surveying after every drill and blast cycle.
After removing blasted material, a photo survey was done and before or parallel to support
processes collateral mapping was carried out, during which all the parameters for rock

PT
classification systems were collected. To create 3D photorealistic models of the rock surfaces
of the excavated tunnel walls, a camera-based photogrammetric method was chosen in

RI
accordance with the adequate resolution and quality for the photo-combined 3D models (Deák
et al., 2014). In the analyzed area of this research, ShapeMetriX3D system (www.3gsm.at)

SC
served as a basis of 3D models (Fig 2a-d). Models were processed and evaluated by the
geotechnical team and after that digital bases were uploaded to the Datamine online 3D
NU
mapping and database system (Advanced Survey) for geological evaluation (Kovács et al.,
2017). Geotechnical survey, field report, and 3D geotechnical advance models have been
MA

collected into this extensive database. Along with systemized orientation data, a number of
other parameters, describing fracture systems, were also recorded. These data were as follows:
type of fractures, joint spacing, planarity, surface roughness, separation, infilling material,
ED

thickness of the infilling material, surface texture.


T

Figure 2a. Tunnel face in class II rock mass (RMR = 73; Q = 2.621; GSI= B3 50-55)
C EP

Figure 2b. Tunnel face in class III rock mass (RMR = 58; Q = 0.451; GSI=B4 40-45)
AC

Figure 2c. Tunnel face in class IV rock mass (RMR = 37; Q = 0.054; GSI=C4 35-40)

Figure 2d. Tunnel face in class V rock mass (RMR = 14; Q = 0.010; GSI= E5 10-15)

GSI values were determined according to results of three quantitative methods and not in
qualitative way with a help of a matrix chart. The geotechnical expert had the chance to
choose between three calculation results (see details below), which were calculated from
geotechnical field mapping parameters – joint orientation, spacing, planarity, roughness, type
ACCEPTED MANUSCRIPT

and thickness of the infill (different joint sets with different orientation and parameters are
presented in Figure 2a-2d with colored lines). The final value was determined by the
geotechnical expert who performed the survey. It is a well known fact that different persons
can describe different rock mass characteristics at the same rock surface (Deák et al. 2014).
Therefore this documentation method was developed during the construction of the first two
repository chambers in 2011 to avoid subjectivity and minimize uncertainty of
characterization. In our practice using GSI quantitative determination methods was a guide for
the experts and limitation for the degree of mischaracterization.

PT
The result contains a letter (a row of GSI matrices) and a number (column of the GSI

RI
matrix). The GSI chart published by Hoek et al. (2013) was used for plotting the results. The
GSI chart and the calculations of Cai & Kaiser (2006) were applied for the verification of the

SC
results. These 268 independent data provide a good representation of the variation of the
Mórágy Granite Formation.
NU
Fig. 3 shows the GSI values determined on site by the geotechnical group in different
tunnels in the chart of Hoek et al. (2013). As also outlined in the figure, the two smaller
MA

tunnels have more scattered GSI values than those of the repository chambers. The repository
chambers were excavated in a rock mass of varying degree of fracturing where the fracture
surfaces are dominantly in fair condition. The western research tunnel was driven in the same
ED

monzogranite rocks as the repository chambers, although it was characterized by better


surface conditions. Research tunnel 3 intersected one of the main shear zones of the site thus
T

it has the widest range of rock mass properties, as shown by Fig 5a in detail. We can conclude
EP

also that most of the GSI values are above 30.


C

Figure 3. Distribution of GSI values from field mapping in different investigated tunnels
AC

2 DETERMINATION METHODS OF GSI VALUES

Most tunnel face documentation carried out around the world were based on the RMR and
Q systems, so the main authors of the GSI (Hoek et al, 1995) already put great emphasis on
the "interoperability" of the various methods when they created the GSI method. In the last
construction phase, since we surveyed the required parameters for the most widespread rock
mass classifications, other approaches were used. Based on the data, we can analyze GSI in a
different way. In this chapter we present the calculation of GSI based on RMR, Q and RMi
methods.
ACCEPTED MANUSCRIPT

Our investigation focused on the seven different equations of GSI determination (Eqs. (1)-
(7)) based on the observations during the excavation of the Bátaapáti radioactive waste
repository.

2.1 RMR-based GSI calculations


In the first attempts to define rock mass geological conditions, Bieniawski’s RMR1989 was
applied in the H-B failure criteria equation (Bieniawski, 1989), which was described as (Hoek
et al., 1995):

PT
GSI1 = RMR1989 – 5 = R1 + R2 + R3 + R4 + R5 (=15) – 5 (1)

RI
where the parameters and their values depend on different conditions.
₋ R1 – Uniaxial Compressive Strength (0-15),

SC
₋ R2 – Rock Quality Designation, RQD (3-20),
₋ R3 – average joint space (5-20),
NU
₋ R4 – joint wall conditions (0-30) and
₋ R5 – water. In the original definition R5 must be defined as dry (i.e. 15).
The RMR1989 index theoretically ranges from a minimum of 0 to a maximum of 100 -
MA

consistent with the GSI scale. It should be stated that GSI=RMR-5 is not anymore suggested
and can not be only applied in certain quality rock masses.
ED

Meanwhile, it was also necessary to determine the relationship between RMR and GSI. On
the basis of several studies, Hoek et al. (2013) suggested the following simple formula for
T

GSI calculation:
EP

GSI2 = 1.5 R4 + 0.5 RQD (2)


C

2.2 Q-value based GSI calculations


AC

The widespread use of Barton’s Q system required the specification of a Q-based method
of GSI determination. In this approach, Jn (joint set number), Jr (joint roughness number) and
Ja (joint alteration number) coefficients are used (Barton, 1995):

 RQD Jr 
GSI3 = 15log    50 (3)
 Jn Ja 

According to Hoek et al. (1995), this equation should be used in following form:

 RQD Jr 
GSI4 = 9ln    44 (4)
 Jn Ja 
ACCEPTED MANUSCRIPT

These equations (Eqs. (3) and (4)) were determined from the relationships based on the
statistical analysis of RMR and Q systems. It is well-known that the RMR value cannot be
calculated exactly from the Q value (Palmström, 2009), up to now several equations have
been published with different constants. The calculation of the RMR value from the Q value
is usually site-dependent (Deak et al., 2012).

Later Hoek et al. (2013) further developed Eq. (3), and suggested the following solution,
applying the Jr, Ja and RQD constants of the Q system:

PT
52 Jr / Ja
GSI5 =  0,5RQD (5)
1  Jr / Ja 

RI
2.3 RMi-based GSI calculations

SC
Not only Hoek and his co-author attempted to provide an increasingly precise solution for
the determination of GSI. Shen and Barton (1997) demonstrated the strong influence of the
NU
size of the rock blocks on the mechanical behavior of the rock mass. One of the most
appropriate methods was suggested by Cai & Kaiser (2006), in which the size of the block
MA

was examined. They applied Palmström’s theory (1995) where jC is the coefficient of joint
conditions, Jp is the volumetric joint count and Vb represents the block volume. In the view
of these coefficients, GSI is:
ED

26,5  8,79 ln Jc  0,9 ln Vb


GSI6 = (6)
1  0,0151ln Jc  0,0253 ln Vb
T
EP

On the basis of these considerations, Russo (2009) suggested the following equation (7):

165
GSI7 = 153  (7)
C

  Jp  0, 44 
1    
AC

  0,19  

During the construction of the chambers and tunnels, we determined GSI considering three
different methods. GSI values were calculated based on Eqs. (2), (5) and (6) right after tunnel
face mapping and the geotechnical engineer gave the appropriate value according to the
results and the experienced rock environment in the tunnel. In our final evaluation of tunnel
face mapping, we also calculated and analyzed the results of the four other equations.
ACCEPTED MANUSCRIPT

2.4 Other methods


Other approaches emphasize and suggest the advantages of digital face mapping and
digital rock mass rating. Poulsen et al. (2015) focused on quantifying the GSI chart to
facilitate the use of the system especially for inexperienced practitioners. Others (refer to
Hong et al. 2017, among others) investigated digital face mapping as a practical tool for the
characterization of rock masses, which can significantly reduce the time required in the field
and helps to avoid exposure to potentially unsafe conditions.

PT
Hong et al. (2017) introduced an approach for the determination which uses fractal
dimension for describing the structure of rock masses and is a simple way to determine rock

RI
surface condition. These two input parameters were the basis of an ANN analysis. In their
conclusion, the GSI value of a jointed rock mass can be quantified by combining image

SC
processing, fractal theory and ANN. However, this method needs further improvements
because it cannot be applied to weaker rock mass.
NU
3 EXAMINATION OF DIFFERENT GSI VALUES
MA

In addition to field data, GSI values were recalculated from the tunnel face mapping
database according to Eqs. (1) - (7). In a further step, data were analyzed in the longitudinal
ED

section of the tunnels. Fig. 4a and 4b show two different rock mass situations. Rock types
were also shown and they also have some correlation with GSI values. The fault zone was
also classified with GSI despite the presence of anisotropy. Although the trends of GSI values
T

are similar, the seven different specification methods resulted in a very wide range of plots.
C EP

Figure 4a. Calculated GSI values along the longitudinal section of research tunnel 3 (Vásárhelyi et al., 2016)
AC

Figure 4b. Calculated GSI values along the longitudinal section of the I-K4 repository chamber

The field mapping values with the mean, minimum, maximum and range values of the GSI
determinations in these tunnel sections are presented in Figs. 5a and 5b. The ranges of the
different GSI systems are between 15 and 38, the average range value is 27. It is 20 in the
most critical zone, thus the calculated GSI values are between 0-20 in the fault zone.
Surprisingly the scatter is the highest directly before and after the fault zone. The real field
mapping values follow the mean not only in trends but also in values. This must be the result
of using different determination methods during field work.
ACCEPTED MANUSCRIPT

Figure 5a. Average GSI values with the minimum/ maximu m values of research tunnel 3

Figure 5b. Average GSI values with the minimum/maximu m values of the I-K4 repository chamber

The calculated minimum and maximum values are presented in Fig. 6. As it can be seen in
the case of weaker rock mass, the difference increases and the correlation is not close to the
origin one. It can be concluded that there is a significant variance between the different

PT
methods. Usually, the smallest values were obtained by using Eq. (7) (theory of Russo, 2009),
while the highest values were obtained from Eqs. (1) and (4). In many cases, the difference

RI
between these two values was above 100 %.

SC
Figure 6. Calculated maximu m GSI values as a function of the calculated minimum GSI values in the
investigated site
NU
4 RELATIONSHIPS BETWEEN SIMILAR METHODS
MA

4.1 RMR-based GSI values

First the RMR-based calculations were compared (Eqs. (1) and (2)). According to Fig. 7.,
the difference between the two calculation methods is not significant if GSI > 60. With
ED

decreasing GSI value, the difference increases. It can be seen that by using the latest method
(Eq. (2)), we can obtain higher GSI values than with the older one (Eq. (1)). The differences
T

between the two values are generally within ±10 compared to each other, similarly to the
EP

results of Bertuzzi et al. (2016). A regression was found between these two calculation
methods:
C

GSI2 = 1.2753 GSI1 + 22,292 (8)


AC

Figure 7. Comparison of the two RMR-based GSI values

4.2 Q-based GSI values


Up to now, 3 equations were developed based on the main parameters of the Q values
(Eqs. 3-5). Eqs. (3) and (4) are used the similar correlation, i.e. there is linear relationship
between the two calculation systems:

GSI3 = 0.724 GSI4 + 18.152 (9)


ACCEPTED MANUSCRIPT

This means that the difference between the two calculations incerases with decreasing GSI
values (the difference is less than 10 if GSI4 >30; and less than 15 if GSI4 > 10):

Δ = -0.2762 GSI3 + 18.152 (10)

where Δ = GSI4 – GSI3

The difference between the two calculation methods is within ±10 compared to each other
in the investigated GSI region. An exponential relationship was found between the method of
Barton (1995) – Eq. (3), GSI3 – and the method of Hoek et al. (2013) – Eq. (5), GSI5 .

PT
GSI5 = 1,6929 e0,0598GSI 3 (11)

RI
SC
Figure 8. Relationship between GSI3 and GSI5

4.3 RMi-based GSI values


NU
There are two calculation methods based on RMi parameters (Eqs. (6) and (7)). The
calculated GSI values are different but strong relationship was found between GSI6 and GSI7 .
MA

This relationship is shown in Fig. 9.

GSI7 = 0.0085 GSI6 2 + 0.744GSI6 – 10,67 (12)


ED

Around GSI values of 50, the results are the same. Under 50, GSI6 values (equation of Cai
& Kaiser, 2006) are always lower than those of GSI7 (equation of Russo, 2009), theoretical
T

differences are dominantly under 10.


EP

Figure 9. Comparison of the GSI values calculated with RMi parameters


C
AC

5 DISCUSSION

The main objective of our research was to analyze the results of the rock mass
classification process in the Hungarian NRWR construction project and determine
relationships between different procedures. Vásárhelyi et al. (2016) analyzed the relationships
between the various methods in detail. In addition to their work, we have analyzed the
documentations performed in other tunnels as well. As a result of our work, field
documentation values of the storage chambers were evaluated separately on sections. In the
longitudinal section of Fig. 4, the changes in GSI values followed the changes in rock type. It
is also important to note that values from the field documentation were found to be close to
ACCEPTED MANUSCRIPT

the mean of calculated values. On this basis, it is justified that several different methods can
be considered for the calculation of GSI. The evaluation showed that there are significant
differences between the seven different procedures. It has been observed that a high degree of
deviation was found not only in case of weaker rock masses.

The calculated GSI values indicate similar trends the basis was the same documentation in
each case. However, significant differences between the minimum and maximum values are
apparent (Vásárhelyi et al., 2016). The difference between the GSI values calculated with

PT
different methods decreases in the case of better rock mass quality and increases in the case of
lower GSI values. It is a wide known observation that other rock mass classification systems

RI
do not work sufficiently in weaker rock masses (GSI<35) and it is possible that the
calculation methods contain some uncertainty comes from this deficiency.

SC
Due to the significant differences, it was necessary to select the acceptable calculation
method, in the case of this project. We used three different GSI determination methods in
NU
field survey, and then obtained an acceptable difference from the mean GSI values of seven
different determination methods. This appears to be an appropriate method of GSI calculation
MA

in the field. The GSI correlation in a similar classification system can be precise, but it can
also give varying results, therefore GSI calculations should be treated very carefully.

The results also showed that there is no consistent equation for the determination of the
ED

Geological Strength Index (GSI) but the GSI determinations based on the same rock mass
characterization method gave similar results.
T
EP

6 CONCLUSION
C

The application of the GSI in a fractured granitic rock environment can be highly uncertain
AC

and different determination methods can result in widely scattered GSI values for the same
rock mass. The use of more than one determination method is recommendable, not only for
inexperienced geotechnical engineers. In our case, three different approaches were applied
continuosly for supplementing the field work. In our opinion, the formulas of Hoek (2013)
and Cai et al (2006) can be used together in variously fractured granite rock environment to
describe the quality of rock masses appropriately. Due to this work, the recent database is
reliable for the next design phase. Therefore, the use of this rock mass characterization
process is recommended for further construction phases as well and it can be an appropriate
reference for other projects.
ACCEPTED MANUSCRIPT

We assume that no corresponding method exists for the calculation and it is necessary to
develop independent systems for all projects. Discrepancies between GSI definitions also
showed that there is still considerable uncertainty about the design of tunnel geometry in a
fractured rock mass.

ACKNOWLEDGEMENT

This paper has been published with the permission of Public Limited Company for

PT
Radioactive Waste Management (PURAM). The project presented in this article is supported
by National Research, Development and Innovation Office – NKFIH 124366 and NKFIH

RI
124508. We would like to thank all our colleagues in RockStudy Ltd. and Mecsekérc Plc. for
their help.

SC
NU
MA
T ED
C EP
AC
ACCEPTED MANUSCRIPT

REFERENCES

Barton, N., 1995. The influence of joint properties in modelling jointed rock masses. Keynote
Lecture, 8. ISRM Congress, 3, 1023–1032.
Barton, N.R., Lien, R., Lunde, J., 1974. Engineering classification of rock masses for the
design of tunnel support. Rock Mech. Rock Eng. 6., 4. 189–236. DOI:
10.1007/BF01239496
Benedek, K., Bőthi, Z., Mező, Gy., Molnár, P., 2009. Compartmented flow at the Bátaapáti
site in Hungary. – Hydrogeology Journal 17, 5, pp. 1219-1232. DOI: 10.1007/s10040-009-
0440-4

PT
Benedek, K., Molnár, P., 2013. Combining structural and hydrogeological data:
Conceptualization of a fracture system. Eng. Geol. 163, 1-10. DOI:
10.1016/j.enggeo.2013.05.018

RI
Bertuzzi, R., Douglas, K., Mostyn, G., 2016. Comparison of quantified and chart GSI for four
rock masses. Eng. Geol., 202, 24-35, DOI: 10.1061/(ASCE)GT.1943-5606.0001644

SC
Bieniawski, Z.T., 1989. Engineering rock mass classification. Wiley, New York
Cai, M., Kaiser, P.K., 2006. Visualization of rock mass classification systems. Geotech. Geol.
NU
Eng. 24, 4., 1089–1102. DOI: 10.1007/s10706-005-7464-x
Carter T.G. 1992. Prediction and uncertainties in geological engineering and rock mass
characterization assessments. Proc. 4th. int. rock mechanics and rock engineering conf.,
MA

Torino, I-1 – I-22


Deák, F., Kovács, L., Vásárhelyi B., 2014. Geotechnical rock mass documentation in the
Bátaapáti radioactive waste repository. Centr. Eur. Geol. 57., 2., 197-211. DOI:
10.1556/CEuGeol.57.2014.2.5
ED

Deák, F., Kovács, L., Vásárhelyi, B., 2012. Comparison of different rock mass classification
at Bátaapáti radioactive waste repository. In: Rock Engineering & Technology for
Sustainable Underground Construction: Eurock 2012, Stockholm, Sweden, pp. 247-259
T

Grocki A.; Schubert W.; Riedmüller G. 2004: New developments for the design and
EP

construction of tunnels in complex rock masses. Int. J. Rock Mech. Mining Sci. 41(3):
497-498. DOI: 10.1016/j.ijrmms.2003.12.033
Hoek, E.; Marinos, P. 2000. Predicting tunnel squeezing problems in weak heterogeneous
C

rock masses, Tunnels and Tunnelling International,32(11) 45-51. .Hoek, E., Brown, E.T.,
AC

1980. Underground excavations in rock. London, Inst. Min. Metall.


Hoek, E., Brown, E.T., 1997. Practical estimates of rock mass strength. Int. J. Rock Mech.
Mining Sci., 34(,8): 1165-1186. DOI: 10.1016/S1365-1609(97)80069-X
Hoek, E., Carter, T.G., Diederichs M.S., 2013. Quantification of the Geological Strength
Index Chart. 47th US Rock Mechanics/Geomechanics Symposium.
Hoek, E., Kaiser, P.K., Bawden, W.F., 1995. Support underground excavations in hard rock.
Balkema.
Hong, K., Han, E., Kang, K., 2017. Determination of geological strength index of jointed rock
mass based on image processing. J. Rock Mech. Geotechn. Eng. 6, 702-708. DOI:
10.1016/j.jrmge.2017.05.001
ACCEPTED MANUSCRIPT

International Atomic Energy Agency, 2003. Scientific and Technical Basis for the Geological
Disposal of Radioactive Wastes. IAEA Technical Report Series Number 413. Vienna.
Kovács, L., Krupa, Á., Schön, R., Gaburi, I., 2017. “Data-Mine” software: Complex earth
scientific documentation of excavations with a uniform, real 3D background. Proc. 20th
Hungarian and 9th Croatian-Hungarian Geomathematical Congress. Pécs, Hungary, 11-13.
May 2017. 60.
Kovács, L., Mészáros, E., Somodi, G., 2015. Rock Mechanical and Geotechnical
Characterization of a Granitic Formation Hosting the Hungarian National Radioactive
Waste Repository at Bátaapáti In: Procs. Eng. Geol. for Soc. Territory – 6., 915-918. DOI:
10.1007/978-3-319-09060-3_166

PT
Kovács, L., Vásárhelyi, B., 2015. Special Requirements for Geotechnical Characterization of
Host Rocks and Designing of a Radioactive Waste Repository. In: Procs. Eng. Geol. for
Soc. Territory – 6., 909-912. DOI: 10.1007/978-3-319-09060-3_165

RI
Marinos V. 2012: Assessing rock mass behaviour for tunneling. Environmental and
Engineering Geoscience, 18 (4): 327-341 DOI: 10.2113/gseegeosci.18.4.327

SC
Marinos, V., 2014. Tunnel behaviour and support associated with the weak rock masses of
flysch. J Rock Mech Geotech Eng 6., 227–239 DOI: 10.1016/j.jrmge.2014.04.003
NU
Marinos, V., 2018. A revised, geotechnical classification GSI system for tectonically
disturbed heterogeneous rock masses, such as flysch. Bull. Eng. Geol. Env., in print. doi:
10.1007/s10064-017-1151-z
MA

Marinos V.; Carter T.G. 2018: Maintaining geological reality in application of GSI for design
of engineering structures in rock. Engng. Geol. 239: 282-297 DOI:
10.1016/j.enggeo.2018.03.022
Marinos, V., Marinos, P., Hoek, E., 2005. The geological strength index – applications and
ED

limitations. Bull Eng Geol Environ, 64.,1., 55–65, DOI: 10.1007/s10064-004-0270-5


Morelli, G.L., 2015. Variability of the GSI Index Estimated From Different Quantitative
Methods. Geotech. Geol. Eng. 33., 4., 983-995. DOI: 10.1007/s10706-017-0384-8
T

Palmstom A.; Stille H. 2007 Ground behaviour and rock engineering tools for underground
EP

excavations Tunnelling and Underground Space Technology, 22(4): 363-376. DOI:


10.1016/j.tust.2006.03.006
Palmström, A., 1995. RMi – a rock mass characterization system for rock engineering
C

purposes. PhD thesis, Univ. Oslo, Norway, 400., www.rockmass.net.


AC

Palmström, A., 2009. Technical note: Combining the RMR, Q, and RMi classification
systems. Tunnel. Underg. Space Techn., 24., 4., 491–492. DOI: 10.1016/j.tust.2008.12.002
Poschl L.; Kleberger J. 2004. Geotechnical risks in rock mass characterization. Tunnels and
Tunnelling International, Part 1—May 2004. pp. 37–9. Part 2—October 2004. pp. 36–8.
Potsch, M., Schubert, W.; Goricki, A.; Steidl, A. (2004): Determination of rock mass
behaviour types, a case study EUROCK2004 and 53th Geomechanics Colloquium. VGE.
Poulsen, B.A., Adhikary, D.P., Elmouttie, M., Wilkins, K.A., 2015. Convergence of synthetic
rock mass modeling and the Hoek-Brown strength criterion. Int. J. Rock Mech. Mining
Sci. 80., 171-180. DOI: 10.1016/j.ijrmms.2015.09.022
Russo, G., 2009. A new rational method for calculating the GSI. Tunn. Undergr. Space
Technol., 24., 103-111. DOI: 10.1016/j.tust.2008.03.002
ACCEPTED MANUSCRIPT

Sari, M., 2015. Incorporating variability and/or uncertainty of rock mass properties into GSI
and RMi systems using Monte Carlo method. In: Lollino G. et al. (ed) Engineering
geology for society and territory. Springer International Publishing, pp 843–849,
doi: 10.1007/978-3-319-09060-3_152
Schubert W. 2004: Basics and application of the austrian guideline for the geomechanical
design of underground structures EUROCK2004 and 53th Geomechanics Colloquium.
VGE (2004)
Shen, B., Barton, N.R., 1997. The disturbed zone around tunnels in jointed rock masses. Int. J.
Rock Mech. Min. Sci. 34., 1., 117-125. DOI: 10.1016/S1365-1609(97)80037-8
Ván, P., Vásárhelyi, B., 2014. Sensitivity analysis of GSI based mechanical parameters of the

PT
rock mass. Periodica Polytechnica – Ser. Civil Eng. 58., 4., 379-386, DOI
10.3311/PPci.7507

RI
Vásárhelyi, B., Kovács, D., 2017. Empirical methods of calculating the mechanical
parameters of the rock mass. Periodica Polytechnica – Ser. Civil Eng. 61., 1., 39-50, DOI
10.3311/PPci.10095

SC
Vásárhelyi, B., Somodi, G., Krupa, Á., Kovács, L., 2016. Determining the Geological
Strength Index, GSI) using different methods Proc.: Rock Mechanics and Rock
Engineering: From the Past to the Future: Eurock 2016. Cappadocia, Turkey, 1049-1054.
NU
MA
T ED
C EP
AC
ACCEPTED MANUSCRIPT

Highlights:

 Different determination methods of the GSI were analyzed


 There are significant variances between different methods
 Calculations based on the same rock mass characterization gave similar results
 The lack of corresponding method cause uncertainty in the use of GSI

PT
RI
SC
NU
MA
T ED
C EP
AC
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9

You might also like